NOAA "Rapid Refresh" Provides More Accurate Forecasts for Aviation and Severe Weather

The Nation relies on numerical weather models, complex computer programs, to process millions of weather observations and produce accurate and timely weather forecasts. Developers must make sure that models are both accurate and fast so that watches and warning can be issued with as much lead time as possible. The Rapid Refresh model will be tested this year to support its deployment as the Nation's operational model for aviation and severe weather forecasting.

The tornado outbreak in the central U.S. on April 7, 2006 demonstrated just how quickly weather conditions can evolve into dangerous storms. Forecasts for severe weather and aviation hazards rely on NOAA computer weather models, which are becoming increasingly more sophisticated and timely. These models are updated every hour over the United States with the latest observations from aircraft, radars, satellites, and weather balloons. The present model, referred to as Rapid Update Cycle, has been developed and improved over the last 12 years by the Forecast Systems Laboratory, now part of the NOAA Earth System Research Laboratory. Rapid Update Cycle is NOAA's most frequently updated numerical weather prediction model. This model is run at the National Weather Service's National Centers for Environmental Prediction (NCEP) in Suitland, Maryland.

Warning area of April 7, 2006 tornado outbreak.

Warning area of April 7, 2006 tornado outbreak.

This year, weather model scientists from NOAA/ESRL and the National Center for Atmospheric Research (NCAR) have again joined forces to develop a new, hourly updated, operational model called Rapid Refresh that is designed to substantially improve the forecasting skill of NOAA. This model will take advantage of the collaboratively developed, state-of-the-art Weather Research and Forecast model referred to as WRF. Rapid Refresh makes use of WRF's optimal combination of numerical and physical parameterizations. The emphasis for the enhanced model is on aviation and severe weather applications in 1 to 12-hour forecasts.

NOAA and NCAR's joint Developmental Testbed Center will conduct core testing of the two models. The purpose of this testing is to compare the two dynamic/numerical techniques of the respective weather research and forecast models. The test results will provide guidance regarding which of the WRF techniques will actually be used in the Rapid Refresh. The configurations selected for these experiments will include 13-kilometer horizontal spatial resolution, cloud fields from the original Rapid Update Cycle, CONUS domain, and 30-day experiments over each of four seasons.

Evaluation Criteria

The evaluation will be based on an overall statistical verification and case study investigations. The results for turbulence, icing, ceiling/visibility and convective weather will be examined over the four seasonal periods, with two versions of physical parameterizations. NOAA agencies and aviation groups, including Federal Aviation Administration (FAA)-sponsored programs, will conduct the examination primarily at NCAR.

The statistical evaluation will include verification against upper-air observations (wind, moisture, temperature), surface observations (temperature, moisture, wind), and precipitation observations. Aviation-specific performance for fields related to icing, clouds (including ceiling), turbulence and convection will also be evaluated.

Shifts in location and timing for precipitation and clouds can be critical for aviation operations and surface transportation. Examples of such shifts are shown in Figure 2, an image of a 12-hour forecast difference for surface temperature valid for February 16, 2006. In this figure, the surface cold front (indicated by the blue shading) is further south in one model than the other.

Surface temperature differences between the existing and proposed models valid at 16 February 2006.

Surface temperature differences between the existing and proposed models valid at 16 February 2006.

Figure 3 illustrates the accumulated precipitation over a 3-hour period and shows a more accurate location for this cold front from the Rapid Refresh.

The precipitation forecast valid at 16 February 2006.

The precipitation forecast valid at 16 February 2006.

Future

Based on these WRF-Rapid Refresh test results, NOAA scientists will make a final recommendation on the dynamic techniques to NCEP this summer. Other partners in the final evaluation will include the following aviation-related users of the current Rapid Update Cycle: the FAA-sponsored Aviation Weather Research Program; the NCEP Aviation Weather and Storm Prediction Center, and specialization teams for turbulence, icing, ceiling/visibility, and convective weather.

Starting this fall, the full Rapid Refresh cycle with a WRF model configuration of the most appropriate dynamic technique and physical parameterizations will be tested at NOAA/ESRL for aviation and severe weather applications. Using the Gridpoint Statistical Interpolation assimilation code, the testing will cover the full domain, including Alaska, Puerto Rico, and all of Canada. Rapid Refresh should replace the Rapid Update Cycle by the summer of 2008, following this and equally intensive testing at NCEP.

The WRF-Rapid Refresh technique test is a significant step, but by no means the only one necessary for this Rapid Refresh implementation. In an effort to go beyond that, NOAA scientists are combining expertise on physical processes, modeling, and data assimilation that will enhance the new model. Their experience in bridging community research efforts to NOAA's operational services will be invaluable for the Rapid Refresh implementation. Again, with this project, NOAA will outdo itself by providing more accurate and timely forecasts of the Nation's weather.

More Information